How do I know if an AI model could work locally on my computer?
AI-assisted, reviewed by our team · Last verified Oct 16, 2026 · How we research
The quickest answer from that thread: LM Studio tells you before downloading whether a model fits your hardware. Rules of thumb from commenters: what matters is the model *file size* compared with your RAM (or GPU memory), not the parameter count, and you need headroom for the context (the document you're asking about). Models larger than your GPU memory still run but slow down as they spill into system RAM. "Mixture-of-experts" models can run faster for their size.
For a contractor laptop: start with a small model in LM Studio or Ollama on whatever machine you own, ask it questions about a spec section, and step up in size only if answers are weak. Apple Silicon Macs share memory between CPU and GPU, which is why they run local models well; oMLX and MLX are built for them. Full sizing table: Which local AI models fit on which hardware.